Ocean Drilling Program Scientific Results Volume 123Gradstein F. M., Ludden, J. N., et al., 1992 Proceedings of the Ocean Drilling Program, Scientific Results, Vol. 123
8. GEOCHEMISTRY OF SEDIMENTS IN THE ARGO ABYSSAL PLAIN AT SITE 765: A CONTINENTAL MARGIN REFERENCE SECTION FOR SEDIMENT RECYCLING
IN SUBDUCTION ZONES
ABSTRACT
Drilling at Site 765 in the Argo Abyssal Plain sampled sediments and oceanic crust adjacent to the Australian margin. Some day, this site will be consumed in the Java Trench. An intensive analytical program was conducted to establish this site as a geochemical reference section forcrustal recycling calculations. About 150 sediment samples from Site 765 were analyzed for major and trace elements. Downhole trends in the sediment analyses agree well with trends in sediment mineralogy, as well as in Al and K logs. The primary signal in the geochemical variability is dilution of a detrital component by both biogenic silica and calcium carbonate. Although significant variations in the nonbiogenic component occur through time, its overall character is similar to nearby Canning Basin shales, which are typical of average post-Archean Australian shales (PAAS). The bulk composition of the hole is calculated using core descriptions to weight the analyses appropriately. However, a remarkably accurate estimate of the bulk composition of the hole can be made simply from PAAS and the average calcium carbonate and aluminum contents of the hole. Most elements can be estimated within 30% in this way. This means that estimating the bulk composition of other sections dominated by detrital and biogenic components may require little analytical effort: calcium carbonate contents, average Al contents, and average shale values can be taken from core descriptions, geochemical logs, and the literature, respectively. Some of the geochemical systematics developed at Site 765 can be extrapolated along the entire Sunda Trench. However, results are general, and Site 765 should serve as a useful reference for estimating the compositions of other continental margin sections approaching trenches around the world (e.g., outboard of the Lesser Antilles, Aegean, and Eolian arcs).
INTRODUCTION
The extent to which the continental crust is recycled back into the mantle via sediment subduction is crucial to our understanding of how Earth's mantle and crust evolved. Primarily, two lines of evidence support sediment subduction. One is based on seismic surveys and drilling that show an absence of accreted sediments in some forearcs (e.g., in the Marianas: Hussong, Uyeda, et al., 1982; Guatemala: Moore, Backman, et al., 1982; and Peru: Warsi et al., 1983). Even where well-developed accretionary wedges occur, often evidence indicates that some sediment is also being subducted beneath décollement structures in the wedge (as in the Lesser Antilles: Westbrook et al., 1988) or in grabens developed in the bending plate (as for the Nazca Plate approaching the Peru-Chile Trench: Schweller et al., 1981). Thus, ample geophys- ical evidence exists to show that sediment is subducted, at least beneath some forearcs.
A second line of evidence for sediment subduction comes from the isotope 10Be. An isotope strongly enriched in soils and marine clays, 10Be is found in measurable quantities in arc volcanics, but not in volcanics from other tectonic settings (Tera et al., 1986). This means that some surface sediments are taken as far as the site of arc magma genesis (-120 km deep). The factors that lead to sediment subduction in some situations, and not others, are still poorly understood (a modern twist on an old soliloquy provides a recent discussion of the various models; Von Huene, 1986).
The issue of how much sediment gets subducted to great depths remains open. Arc magmas incorporate some quantity of sediment
1 Gradstein, F. M., Ludden, J. N., et al., 1992. Proc. ODP, Sci. Results, 123: College Station, TX (Ocean Drilling Program).
2 Lamont-Doherty Geological Observatory and Department of Geological Sciences of Columbia University, Palisades, NY 10964, U.S.A.
3 Département de Géologie, Université de Montreal, C.P. 6128 Succ. A, Mon- treal, Quebec, H3C 3J7, Canada.
and provide our best means for estimating the fluxes involved. However, these calculations require knowledge of the geochemi- cal characteristics of both the influx (sediment and crust ap- proaching trenches) and the output (arc volcanics). Although a fairly comprehensive global data base exists of the geochemical composition of arc volcanics, a method has yet to be developed for estimating the composition of the diverse sediment sections approaching trenches. A considerable amount of geochemical data exist for marine sediments. Most chemical analyses of sedi- ments, however, consist of a few elements specific to oceano- graphic problems, and not necessarily solid earth ones. These data provide a first-order understanding of the systematics of sediment compositions, but do not constrain well what compositions are appropriate for individual subduction zones. Estimates of sedi- ment compositions outboard of trenches usually are based on analyses of a few surface sediments, whole-ocean averages, or average "pelagic sediment." These estimates need to be refined. For example, elements that are used as tracers for sediment in- fluxes to arc magma sources (e.g., K2O; Karig and Kay, 1980) may vary considerably in "pelagic sediments," on the order of all igneous rocks on the face of Earth. This is because sediments represent mixtures of biogenic carbonate or silica, which are devoid of most trace elements, and continental detritus or Fe-Mn oxides, which are rich in many trace metals. Thus, although ratios of certain elements and isotopic compositions of sediments are fairly well constrained, the actual element abundances may vary enormously. Ultimately, it is concentration data, not element or isotopic ratios, that are necessary for answering the question of how much!
Here, we discuss the geochemistry of the sedimentary section of Site 765, drilled during Leg 123 in the Argo Abyssal Plain, south of Java. These sediments may someday be consumed in the Sunda Trench. Full characterization of potential crustal fluxes into the Sunda Trench should include other sedimentary compo- nents that may be added to Site 765 as it drifts northward: volcanic ash derived from the arc itself and clastic material derived from
167
T. PLANK, J. N. LUDDEN
the Ganges-Brahmaputra river system via the Nicobar Fan. In addition to sedimentary components, the chemical additions to the basaltic substrate via seawater alteration constitutes another crus- tal component. Gillis et al. (this volume) present some preliminary conclusions regarding the geochemical fluxes involved during alteration of the basaltic crust at Site 765.
Although the data presented here have obvious bearing on problems specific to the Sunda Arc as well as to the sedimen- tological history of the Australian margin, the thrust of this study is more general: to provide a reference data set for sediment subduction globally. Indeed, Site 765 represents an end-member of sorts. The site is situated adjacent to a passive margin, and so represents the crust first formed when an ocean basin opens, and the last consumed when an ocean basin disappears down a sub- duction zone. The sediments that have accumulated at Site 765 are dominated by detrital material derived from the Australian continent. Thus, Site 765 should serve as a reference site for other subduction zones proximal to continental sediment sources (e.g., the Lesser Antilles, the American, and the Eolian/Aegean arcs). We hope to provide here a methodology for constraining sediment influx to this class of subduction zones, and in doing so, to begin to answer the question of how much?
First, we present a geochemical stratigraphy of Site 765 sedi- ments, attempt to tie the geochemical variability to the lithologic variability, and briefly speculate about the provenance of the sediments. Next, we devise a method for estimating the bulk composition of the entire sediment section. Finally, we discuss the relevance of this site to sedimentary sections globally, as well as regionally along the Sunda Arc.
GEOLOGIC BACKGROUND
Site 765 is situated in the Argo Abyssal Plain, a triangular region of some of the oldest crust of the Indian Ocean, sandwiched between the northwestern margin of Australia and the Java Trench (see Fig. 1). The crust drilled at Site 765 represents the first oceanic crust formed during rifting of the Australian margin in the Earliest Cretaceous. This site is now only 500 km away from being consumed at the Sunda Trench, south of Java. The sedimen- tary section that has accumulated is thin for a passive margin sequence owing to the arid climate and low relief of western Australia. Although several active and explosive volcanoes are located on Java and the Lesser Sunda Islands, Site 765 has yet to enter the region of extensive ash falls determined by Ninkovich (1979). Thus, the ultimate source of much of the sediment at Site 765 is the Pilbara and Kimberly blocks of northwestern Australia craton, which are Archean to Proterozoic in age (Fig. 1). Although Site 765 is at abyssal depths, sedimentation rates have been higher than typical abyssal rates (see Ludden, Gradstein, et al., 1990), especially during the Neogene (averaging 27 m/m.y.) due to the continual supply of material from the Australian Shelf and Exmouth Plateau via turbidity flows. Even though well below the calcite compensation depth, the hole has a high carbonate content owing to the rapid influx and burial of pelagic carbonates from the Australian margin.
At Site 765, roughly 950 m of sediment was cored above the basaltic basement. The sedimentary section can be divided in two, corresponding grossly to the Cenozoic and Cretaceous sections. The Cenozoic section is dominated by calcareous turbidites, prob-
IIO°E Figure 1. Location map of Site 765 with simplified geology of northwestern Australia. Water depth contours in meters.
168
GEOCHEMISTRY REFERENCE SECTION FOR SEDIMENT RECYCLING
ably originating on the Exmouth Plateau and fed by the Swan Canyon (Fig. 1). Although calcareous sequences also occur throughout the Cretaceous section, this lower section is domi- nated by pelagic clays. Within this simple division, however, tremendous lithologic diversity is represented by foraminiferal sands, nannofossil oozes, radiolarites, Mn-rich horizons, and red, green, and black clays.
SAMPLING AND ANALYTICAL DETAILS Our sampling strategy was to take one 40-cm3 sample at each
core that was representative of a dominant lithology in that core. Thus, we have a fairly evenly spaced sampling every 10 m or so down the entire 930-m section. Some sections of cores were subsampled (three individual turbidite units and red and green clay units in the lower half of the hole), and several samples were taken in adjacent intervals to assess variability at the centimeter scale. Samples were powdered and homogenized by first baking to 110°C, then pulverizing in either a tungsten carbide shatterbox or an alumina ball-mill.
About 70 samples were analyzed on board the Resolution using X-ray fluorescence (XRF) for all the major elements and for Rb, Sr, Ba, Y, Zr, Nb, V, Cr, Ni, Cu, and Zn (Table 1). Another 70 samples were analyzed at Lamont-Doherty (LDGO) by direct-cur- rent plasma emission spectrometry (DCP) for all the major ele- ments and for Sr, Ba, Y, Zr, V, Cr, Ni, Cu, Zn, and Sc (Table 1). Forty samples were selected for additional instrumental neutron activation analysis (INAA) at the Université de Montreal, for REE, U, Th, Sc, Cr, Hf, Ta, W, Co, As, Sb, Ba, and Cs (Table 2). The major- and trace-element analyses for these same samples or for adjacent intervals are presented in Table 1. Details of the XRF and INAA procedures are given in Ludden, Gradstein, et al. (1990) and in Francis and Ludden (1990), respectively. Analysis of sediment samples by DCP required new procedures, which are described below.
Different routines were set up for running clay- and carbonate- rich samples by DCP. The clay-rich samples (with 10% CaO) proved as straightforward to analyze as igneous rocks, and proce- dures were followed similar to those outlined in Klein et al. (1991). For each batch of clay samples (usually 10 unknowns), the USGS standards SCO-1 and QLO, as well as an in-house Aleutian andesite standard, LUM-37, were used to establish the calibration curves.
The analysis of carbonate-rich samples (>10% CaO) required some modifications to our routine method. Preliminary shipboard work revealed a problem in alkali loss upon ignition. We over- came this problem on board the ship by analyzing the alkali elements (K, Na, and Rb) on unignited powder pellets, although these measurements are inherently less accurate for K and Na, which are normally analyzed on fused glass disks. A new proce- dure had to be developed for the DCP method as well, because samples typically are fused and dissolved after ignition. However, comparison with analyses by total HF-HCIO4 dissolution indi- cated that alkali loss occurred only during ignition of samples (30 min at 1000°C), not during LiBθ2 fusion. In the most carbonate- rich samples, some K was lost during the standard 15-min fusion at 1050°C; times were then reduced to 5 min, which seemed sufficient for fusion without alkali loss. Thus, carbonate samples were fused and dissolved without first oxidizing or devolatilizing. Total volatile loss on ignition (LOI) was determined for separate splits.
Significant Ca interferences or enhancements occurred on the Al line and on all the trace-element lines of our multi-element cassettes (the specific wavelengths used for each element are available upon request). With the exceptions of Sc, Y, and Zr, however, the Ca interferences are linear and easily corrected by running a pure CaO standard during each run. Calibration curves
for each batch of unknowns were constructed using pure CaO, LUM-37, and mixtures of the two in the proportions 1:1 and 1:3. These CaO-andesite mixtures provided us both with standards that closely resembled the unknowns and with dependable values for the trace elements, which is important because few well-char- acterized carbonate standards exist. Sc, Y, and Zr have large Ca interferences and are matrix sensitive (Ca enhances and Si sup- presses). As a consequence, the standards do not form good calibration curves, and the Sc, Y, and Zr data are not accurate (10%-20% relative). More recent tests have shown that matrix problems are reduced by using a factor of 2 greater dilution (1:500) and a greater flux to sample ratio (10:1). With this proce- dure, standards form acceptable lines, and thus the Sc, Y, and Zr data are more accurate (5%), but because of the greater dilution, the peak-to-background ratios suffer, and the data are less precise (10%).
XRF analyses were presented in Ludden, Gradstein, et al. (1990), but have been reproduced here (Table 1) because an additional normalization factor was applied. In addition, K2O was re-run for the powder pellets by XRF at the Université de Mon- treal, and these newer analyses are reported here. On board the ship, the standard SCO-1 was run with every batch of samples; the precision based on these replicates is good (generally better than 2% relative for the major elements and 4% for the trace elements; see Table 21 in Ludden, Gradstein, et al., 1990). The DCP data are similarly precise, based on analysis of an Indian Ocean brown clay sample (IOBC) that was used to monitor drift during 10 different runs (Table 3). The Na2θ and MnO values determined by DCP are more precise than the XRF determina- tions, while the Zr and Y XRF values are preferred because of the matrix problems with the DCP mentioned above.
Although both methods are precise, the agreement between the two varies for different elements. Powders analyzed using both XRF and DCP show consistent discrepancies. These differences are almost always of the same direction and magnitude as the differences between the accepted values for SCO-1 (Govindaraju, 1989) and those determined by XRF. Thus, normalization factors were applied to the XRF data based on the DCP duplicates and the accepted SCO-1 values (see Table 3 for the original XRF average for SCO-1, the Govindaraju values, and the values after normalization). The XRF Ce values and the DCP Sc values were also adjusted to agree with the more precise INAA values. Thus, the data presented in Table 1 show minimal analytical biases (all elements generally agree among methods to 5% relative).
Analyses of adjacent samples, where one was powdered in the ball mill and one was powdered in the shatterbox, indicate con- tamination of a few elements by the tungsten carbide shatterbox (2-4 ppm Co, 0.3-0.4 ppm Ta, and 35-60 ppm W). Aluminum contamination caused by powdering in the alumina ball mill, however, was negligible. The powdering method is listed along with the INAA analyses in Table 2.
GEOCHEMICAL VARIABILITY
The simplest way to estimate the bulk composition of a sedi- mentary section might be to sample continuously down the core, analyze the samples, and then average them. This is impractical for a section that is almost 1 km long, such as was cored at Site 765. This is certainly an impractical method for estimating the flux of material entering oceanic trenches globally. Because our sampling and analytical efforts were intensive for Site 765, we can estimate fairly accurately its bulk composition simply by averaging the analyses reported in Tables 1 and 2. However, our aim was not only to calculate the bulk composition of Site 765 sediments, but also to develop a less analytically intensive method for estimating sections elsewhere. One advantage to working with DSDP/ODP cores is the wealth of lithological and mineralogical
169
Table 1. Analyses of major and trace elements in Site 765 sediments. Table 1 (continued).
© 123-765B
Ce
Ba
Sc
XRF
30.97 .67
40.16 .02
31
XRF
32.39 .27
110.2 95
34.5 52
22.5 1033
4H-3, 145-150
15.02 5.63 .330 3.50 2.21 4.01 2.65 .166 11.5 3.42
.82
14.51 4.86 .131 2.99 3.63 3.90 2.47 .155 12.5 5.41
.98
9.8
41.55 .09
39.29 .26
XRF
39.40 .44
29
XRF
26.46 1.40 1.31 .142
.67
30.73 1.08 1.23 .186
39
34.74 .94
1.13 .143
XRF
33.40 .64
.29
XRF
39.98 .35
XRF
XRF
34.49 .77
XRF
32.88 .65
1.29 .113
30.60 58.48
26
XRF
34.36 .27
XRF
32.30 .62
1.31 .123
29.80 57.48
XRF
34.17 .75
1.17 .156
30.59 60.00
XRF
32.62 .46
1.23 .106
29.50 57.23
21
Clay XRF
54.56 .971
DCP
32.25 1.09 1.15 .113
XRF
36.00 .66
1.31 .109
32.73 63.47
18
DCP
31.98 1.10 1.15 .100
34.84 .68
DCP
32.84 1.10 1.15 .100
XRF
.92
17
XRF
.48
123-765B
10.04
36.22 .32
DCP
39.39 .71
40.24 .33
.68
40.97 .13
45.65 .53
38.73 .80
11.18
43.21 .52
41.10 .17
XRF
41.47 .25
DCP
39.71 .91
36.53 .25
1.08 .070
34.00 68.81
15.27
XRF
42.11 .77
41.71 .68
DCP
15.44
36.41 .57
12.44
22.54 1.11
27.11 .70
45.88 .41
mö 9
ts> 123-765C
Depth: Color:
Lithology: Method:
SiO2 TiO2
AI2O3 FeO
Ce Ba Sc
Na2O K2O
P2O5 LOI
Ce Ba Sc
46.38 .38 .34
30.37 .72 .94
.084 27.68 54.23
39 56.6
47.22 .36
32.88 .93
XRF
45.40 .17 .40
.051 38.04 81.80
13 28.8
86.5 36.3 666
XRF
XRF
Clay XRF
62.24 .601
14.49 5.43 .059 3.87 1.82 2.28 2.12 .141 6.94 2.42
.04
.88
.18
.67
198.4 50.5 136
18.54 8.54 .049 2.96 1.54 1.72 2.14 .097 8.99 4.08
.00
.67
.00
.70 45.68
Ce Ba Sc
Na2O K2O
P2O5 LOI
Ce Ba Sc
146.4 24.7 166
175.7 70.4 181
DCP
144.2 31.4 207
.02
.67
.03
145.2 30.6 210
Clayst. Cc. Clayst XRF
.17
.01
.17
.05
131.3 26.4 262
7.9 5.33
97.0 19.9 201
DCP
12.41 1.09
.95 .085
.33
.13
123-765C
Na2O K2O
P2O5 LOI
Ce Ba Sc
Na2O K2O
P2O5 LOI
Ce Ba Sc
XRF
XRF XRF
44R-2, 38-42
.89 2.91 1.16
19.66 1.18 1.45 .140
12.44 7.22 .180 2.95 1.07 1.75 2.49 .165 5.19 1.00
.04
11.44 1.40 1.82 .160
.50
.22
.58 1.24 1.66 .100 4.23
.58
.15
138.3 31.2 100
123.0 20.7 88
818 15.0
123-765C 44R-4, 45R-2, 47R-5, 50R-1, 51R-1, 55-59 70-75 46-55 110-113 84-86
Depth: 763.55 770.40 793.86 816.70 825.84 Color: Ol gray Dusky red Gy brown Brown Dk gray
Lithology: Clayst. Clayst. Clayst. Clayst. Clayst. Method: XRF XRF DCP XRF XRF
52R-2, 110-118 837.10
55R-4, 2-8
867.52 Br gray
Clayst. Rad clayst Rad ooze Rad clayst DCP DCP DCP DCP
69.14 .700
.29
.67
.09
122.1 22.6
.62 1.31 1.95 .099 4.64
.58
.01
.56 1.24 1.96 .100 4.00
.42
.18
10.4
81.7 28.9 127
57.6 79.6 101
.75 1.16 1.97 .156 4.20
101.9 26.5 113
.49
DCP
120.9 19.7
Clayst. DCP
72.49 .515 9.03 6.00 .644 2.50 1.04 1.23 1.63 .212 4.70
134.6 37.7 190
5961 11.4
123-765C 58R-4, 59R-4, 60R-5, 60R-5, 61R-4, 61R-5, 61R-5, 62R-3, 62R-3. 62R-4, 67-71 39-45 120-123 120-123 92-94 81-85 81-85 73-79 80-84 19-21
Depth: 896.37 902.79 914.70 917.00 922.32 923.71 923.71 930.13 930.20 931.09 Color: Brown Gy red Brown Brown Brown Gy brown Gy brown Gy brown Gy brown Gy brown
Lithology: Clayst. Clayst. Ash Ash Ash Cc.Clayst Cc.Clayst Clayst. Clayst. Clayst. Method: XRF DCP DCP XRF XRF XRF DCP DCP XRF DCP
SiO2 TiO2
Na2O K2O
P2O5 LOI
Ce Ba Sc
.68 1.22 1.55 .156 4.41
.33
.01
.56 .075 6.22
168.0 8.6 136
158.4 32.7 225
10.67 9.58
142.2 30.4 139
.50
.00
143.0 55.4
38.0 490
135 16.4
Oxides in wt%; elements in ppm. Abbrev: light (It), dark (dk), olive (ol), gray (gy), green (gr), brown (br), black (bl), yellow (yell), calcareous (cc), claystone (clayst), siltstone (siltst), dolomitic (dol), radiolarian (rad).
O
Table 2. Instrumental neutron activation analyses (EVAA) of Site 765 sediments.
123-765B Depth: Color:
Cs Ba Sc Hf Ta W Cr
Co As Sb Th
1.93 842
5.10 1.45 .297 1.21 31.7 6.36 2.19 .566 3.77 1.52
4H-6-23 36.23
48.68 16.94
10.387 10.81
3.65 719
8.74 2.21 .442 2.21 53.8 9.13 2.92 .817 6.05 3.14
10H-5-89 93.29
we 17.6 30 7 13.2 2.78 .582 .461 1.59 .245
3.00 427
16H-1-42 144.62
we 17.6 30.7 13.2 2.78 .582 .461 1.59 .245
3.00 427
16H-1-42 144.62
we 17.6 30.7 13.2 2.78 .582 .461 1.59 .245
3.00 427
16H-1-42 144.62
we 17.6 30.7 13.2 2.78 .582 .461 1.59 .245
3.00 427 7.41 2.94 .883
65.57 42.1
16H-1-50 144.70
Al
17.5 30.8 13.4 2.93 .550 .404 1.58 .255
3.10 468 7.58 2.71 .483 6.98 41.2 6.72 3.45 .426 6.39 1.53
17H-5-56 160.46
Al
18H-1-13 163.73 Gy olive
Cs Ba Sc Hf Ta W Cr
Co As Sb Th U
22X-1-88 203.28 Ol gray
2.10 2.78 .218 1.00 20.2 2.74 3.54 .436 2.45 2.22
26X-1-63 241.83
Al
39X-1-20 366.90 Gy olive Dol. clay
Al
4.04 264
11.10 3.54 .615 2.22 93.1 9.13 3.65 .600 7.75 1.44
T . PLA
Cs Ba Sc Hf Ta W Cr
Co As Sb Th
7.10 243
5.55 204
1.78 112
4.19 1.35 .257 2.69 34.8 7.16 3.53 .350 2.80 1.54
11R-1-133 447.33
.201
11R-4-61 451.11 Gy blue
1.450 .996 3.47 .567
1.450 .996 3.47 .567
17R-2-60 504.50
3.79 206
11.41 9.11 .551 5.16
4.91 477
17.2 30.4 12.3 2.70 .519 .380 1.40 .1.97
1.96 1138 4.86 1.32 .324 1.35 26.6 8.68 5.24 .317 5.12
.36
Cs Ba Sc Hf Ta W Cr
Co As Sb Th U
25R-1-92 579.92
1.39 3138 5.39
Clay WC
6.32 2149
22.75 5.35
Cs Ba Sc Hf Ta W Cr
Co As Sb Th U
39R-3-101 715.71 Ol gray Clayst.
WC
6.32 2149
22.75 5.35
WC
5.57 222
26R-4-42 593.22 Br gray
37.25 5.59 .678 9.78
WC
29.05 33.3
20.22 1.26 .690 8.69
27.97 13.39
.817 9.89
35.58 39.67 1.020 13.08 2.17
62R-3-80 930.20
5.12 1473
16.67 5.42
1.179 40.43
2.06 363
16.53 .405 3.66 1.48
5.12 1473 16.67 5.42
12.56 39.5
.50
All elements in ppm. See Table 1 for major and trace element analyses for samples or adjacent intervals. Powdering method: tungsten carbide shatterbox (WC), alumina ball mill (Al). Other abbreviations as in Table 1.
T. PLANK, J. N. LUDDEN
Table 3. Precision and accuracy of DCP and XRF analyses.
SiO2
TiO2
AI2O3
99.19
89 2 0 7
3 5 8 24
4.64 4.44 1.24
1.48 2.12
SCO-1 XRF
68.22 .750
13.0 1 7 6
57.0 5 4 3
11.2 163
1.049 3.00 .239
51.2 6 1 3
Oxides in wt%; elements in ppm.
data that are routinely reported (smear slides, XRD analyses, visual core descriptions, etc.). If geochemical variability can be tied to lithologic variability, then potentially accurate estimates of sediment sections might be made, based only on published core descriptions and a few chemical analyses. Thus, an important first step is to examine the relationship between the geochemical and the lithological variations in Site 765 sediments.
Dilution
The first-order control for variability of almost all the elements analyzed in Site 765 sediments is dilution by calcium carbonate
(cc) (Ludden, Gradstein, et al., 1990). An element such as Al, for instance, is quantitatively diluted by cc. Figure 2 shows that as cc varies from 0 to 100 wt%, AI2O3 varies from about 20 to 0 wt%. Cc is present in Site 765 sediments largely as nannoplankton and foraminifer skeletons that were transported to the site via turbidity flows. A typical calcareous turbidite consists of a foraminiferal sand at the base (almost 100% cc), and a long interval of feature- less olive nannofossil ooze that grades upward to a white nanno- fossil ooze that is often bioturbated (see Dumoulin et al., this volume). Intervals between turbidites typically consist of green clay, with essentially no cc. Because the cc content may vary from 0% to almost 100% in a single turbidite sequence (including the clay-rich interval), the entire compositional range of the hole for an element such as Al can be observed on the centimeter to meter scale of a single graded sequence. Calcareous lithologies are much less common in the Cretaceous section, and so a first-order depth variation in most elements is lower concentrations in the Cenozoic section from dilution by calcareous turbidites and an increase with depth from the predominance of clay-rich lithologies.
A few elements (Sr, U, Ba, and P) do not show cc dilution relationships because they take part in the biologic cycles of the oceans. For example, U and Sr are taken into the carbonate shells of marine organisms. Ba and P are often enriched below zones of high biological productivity (Schmitz, 1987; Toyoda et al., 1990, and references therein). Mn exhibits a complex distribution that is not related to cc dilution, but may reflect post-burial mobility from its redox chemistry (Compton, this volume).
Although a large part of the variation in most element concen- trations results from cc dilution, significant variation at low cc contents also exists. Examination of just the low cc samples shows another dilution effect caused by silica (Fig. 2). Thus, almost all of the AI2O3 variation in Site 765 sediments may be explained by dilution of an end-member with about 20% AI2O3, by cc and silica. The excess silica reflects either radiolarian-rich intervals or detrital quartz. Figures 3A and 3C show good agreement between Siθ2/Al2θ3 (which is insensitive to cc dilution) and radiolarian abundances in the sediments. The scattered high Si2θ/Al2θ3 values in the upper 500 m of the section mark coarse- grained bases of turbidites, where detrital quartz and heavy min- erals may concentrate (Zr and Ti also may be enriched in these
2 0 -
^ 1 5 -
CaCO3 (wt%)
SiO2 (wt%)
Figure 2. AI2O3 vs. CaCθ3, and AI2O3 vs. Siθ2 for a subset of low carbonate samples (<6 wt% CaO)
that exhibit dilution effects on Site 765 sediments.
176
ε Q- a.
Depth (mbsf)
Figure 3. Dependence of Siθ2 and Ba on radiolarian abundances at Site 765. A. Siθ2/Ahθ3 (wt% ratio) of sediments vs. depth. Open triangles represent turbidite bases enriched in detrital quartz. B. Ba contents of sediments vs. depth. C. Radiolarian abundances in sediments vs. depth (from Table 9 in Ludden, Gradstein, et al., 1990). All three variables are high in the upper 100 m and lower 600 m of Site 765.
intervals). Sediments having high abundances of siliceous fossils often have high Ba contents (Schmitz, 1987, and references therein). The highest Ba contents in Site 765 sediments (up to 10,000 ppm) occur in the radiolarian-rich interval between 800 and 900 mbsf, and a rough correspondence exists between Ba and radiolarian distributions (Figs. 3B and 3C).
Although the largest control on concentrations of elements in Site 765 sediments is cc and silica dilution, some elements vary significantly even after removing the effects of dilution. Most of the dilution signal can be removed by normalizing element con- centrations to Al, because Al is quantitatively diluted by cc and silica. In addition, Al is associated almost exclusively with the detrital phase in sediments; thus elements that form constant ratios with Al are diagnostic of the composition of the detrital
phase. In a later section, we will discuss the provenance of the detrital phase in Site 765 sediments. First, however, we will discuss those elements that do not form constant ratios with Al, but instead reflect changes in sediment mineralogy or lithology down the hole.
Mg, Cr, and Sr in Diagenetic Clays and Carbonates
Figures 4 and 5 show anomalously high MgO/Al2θ3,0/AI2O3 and Sr values in the interval between 200-400 mbsf. This interval contains an unusual mineral association: aragonite, dolomite, and the magnesian clay minerals, palygorskite and sepiolite (Compton and Locker, this volume). Because of the intimate intergrowths of
1.6
Depth (mbsf)
800 1000
Figure 4. Enrichment of Mg and Cr in magnesian clay minerals at 200 to 400 mbsf at Site 765. A. MgO/Ahθ3 (wt% ratio) in sediments. B. Cr (ppm)/Ahθ3 (wt%) in sediments. C. Proportion of magnesian clays (palygorskite, sepiolite, and I/S/C) relative to total clays, as estimated by Compton and Locker (this volume).
177
3000"
Depth (mbsf)
800 1000
Figure 5. A. Sr contents of Site 765 sediments. B. Aragonite abundances from Compton and Locker (this volume). Sr is enriched in aragonite relative to clays and calcite.
dolomite and fragile palygorskite fibers, Compton and Locker (this volume) favor a diagenetic origin for the magnesian clay minerals. The bulk composition of the sediments reflects this mineral association: Sr is high because aragonite is enriched in Sr over calcite (Figs. 5A, 5B); MgO is high because of the presence of dolomite and the magnesian clays (Figs. 4A, 4C). Compton and Locker (this volume) suggest that diffusion from seawater sup- plies the Mg required to form the diagenetic dolomite and clay minerals. The high Cr/AhCb values (Fig. 4B) are less easily explained; either the source of the sediments in this interval was enriched in Cr, or the diagenetic reactions that led to the formation of the magnesian clays favored Cr enrichment. We are unaware of any published accounts of Cr-rich varieties of palygorskite or sepiolite.
Potassium and Clay Minerals
K2O/AI2O3 varies by more than a factor of two in Site 765 sediments (Fig. 6A), and variations in this ratio appear to reflect variations in the clay mineralogy (after Compton and Locker, this volume). The K2O content of the bulk sediments varies roughly with the percentage of illite (a high K clay mineral) relative to the K-barren clay minerals, kaolinite and smectite (Figs. 6A, 6B). The low K2O/AI2O3 values around 400 m reflect dominance of kao- linite, while the peak around 600 m reflects high illite and K-feld- spar contents.
Iron, Manganese, and Cretaceous Clays
Both FeO/Al2θ3 and Mn/AhCb increase dramatically at about 600 mbsf (Figs. 7A, 7B), which corresponds roughly with the Cretaceous/Tertiary boundary. The Cretaceous section at Site 765
0.3"
CO
0.0 200 400 600
Depth (mbsf) 800 1000
Figure 6. Dependence of K2O content of Site 765 sediments on clay mineral-
ogy. A. K2O/AI2O3 (wt% ratio) of sediments. B. Proportion of illite relative to
sum of illite, kaolinite, and smectite (I/S <10%) from Compton and Locker
(this volume).
is characterized by a decrease in abundance of cc turbidites, a decrease in sedimentation rate, an increase in amount of pelagic clay, and an increase in (volcanogenic) smectite (Ludden, Grad- stein, et al., 1990). Several of these factors may have led to an increase in the Fe and Mn contents of these sediments. The smectite that dominates the lower 600 m is Fe-rich and Al-poor relative to the detrital illite and kaolinite that dominate the upper 600 m (Compton and Locker, this volume). Thus, the increase in abundance of smectite in the lower section is partly responsible for the increase in FeO/Al2θ3. The slower sedimentation rates in the Cretaceous section, in part reflecting a decrease in rapidly deposited turbidite sequences, favors enrichment and preser- vation of the Fe-Mn oxyhydroxides that constantly rain onto the seafloor. Thus, Cretaceous sediments have a higher hydrogenous component and higher Fe and Mn contents. Finally, FeO and MnO are anomalously high in the clays immediately overlying the basaltic basement (Figs. 7A, 7B), suggesting a ridge hydrothermal origin for some of the Fe and Mn.
REE Abundances, Patterns, and Ce Anomalies
Like Fe and Mn, the rare earth elements (REE) increase dramatically in the lower 600 m (Fig. 8A). A higher hydrogenous component in Cretaceous sediments might have led to higher REE contents because Fe-Mn oxyhydroxide floes scavenge REEs from the water column (Aplin, 1984, and references therein) and may enrich underlying sediments. However, Sm correlates well with
178
Depth (mbsf)
Figure 7. Increasing FeO and Mn in the lower 600 m of Site 765. A. FeO/Ahθ3 (wt% ratio) in sediments. B. Mn (ppm)/Ahθ3 (wt%) in sediments.
P2O5 in Site 765 clays, and the correlation is similar to that of recent Pacific pelagic clays (Fig. 9). One might thus speculate that phosphate phases, such as fish teeth and bones, exert a dominant control on abundances of the REEs in Site 765 sediments, similar to what has been suggested for recent Pacific pelagic sediments (Toyoda et al., 1990). Low sedimentation rates and high biologi- cal activity may have led to higher biogenic phosphate contents in pelagic clays (Toyoda et al., 1990).
The chondrite- or shale-normalized REE patterns also exhibit differences between Cenozoic and Cretaceous samples (Figs. 10A and 10B). The REE patterns of the Cretaceous sediments are typically enriched in the middle REEs, producing concave down- ward shale-normalized REE patterns (Fig. 10B) and high Sm/Yb contents (Fig. 8B). The hydrogenous and phosphate phases, as well as the volcanic source of the smectites in the Cretaceous sediments, may all contribute different REE patterns, and Sm/Yb does not correlate simply with any of these components.
Ce anomalies form in the marine environment as a result of the contrasting behavior of oxidized Ce 4 + relative to the other domi- nantly trivalent REEs. Fe-Mn floes preferentially scavenge Ce 4 +
from seawater, which leads to positive Ce anomalies in Mn nodules and hydrogenous Fe-Mn crusts and a negative anomaly in seawater (Elderfield and Greaves, 1981, and references therein). Site 765 sediments exhibit both positive and negative Ce anomalies, with positive anomalies more common in Cretaceous sediments (Fig. 8C). The magnitude of the Ce anomaly is consis- tent within three different sediment types (Fig. 11). First, samples having the largest positive Ce anomalies are high in Mn, reflecting
GEOCHEMISTRY REFERENCE SECTION FOR SEDIMENT RECYCLING
1.2
1.0-
0.5-
0.0
A.
Depth (mbsf)
Figure 8. Variations in the REEs with depth at Site 765. Dashed line indicates approximate position of the Cretaceous/Tertiary boundary. A. Sm (ppm)/Ahθ3 (wt%) is higher in Cretaceous sediments. B. Sm/Yb (ppm ratio) shows greater heavy REE-depletion in Cretaceous sediments. C. Ce anomaly reflects Ce deviation from La and Nd in chondrite-normalized patterns and may be calculated from [3Cen/(2La« = n + Ndn)], where n indicates chondrite-normal- ized concentrations. Ce anomalies are generally negative (<l.O) in Cenozoic samples and are more commonly positive (>l.O) in Cretaceous sediments.
a significant hydrogenous component that has scavenged REEs from seawater. Second, cc-rich sediments have negative Ce anomalies. Foraminifer tests, although almost devoid of REEs themselves, may become coated by an authigenic Fe-Mn oxide phase rich in REEs (Palmer, 1985). These oxide phases have been
179
40
•
0.8 1.0
Figure 9. Sm and P2O5 in recent Pacific pelagic clays (from Toyoda et al., 1990) and Site 765 clays (<6.0 wt% CaO).
shown to possess negative Ce anomalies inherited from bottom water (Palmer, 1985). Finally, clay-rich samples have Ce anoma- lies that depend upon their P2O5 contents, showing a relationship identical to recent Pacific pelagic sediments (Toyoda et al., 1990). Phosphate phases (such as fish teeth) may have large negative Ce anomalies (Elderfield and Pagett, 1986, and references therein), and so increasing amounts of phosphate may lead to larger nega- tive Ce anomalies.
This analysis of the REEs in Site 765 sediments suggests that downhole variations in abundances of REEs and their patterns are dependent upon downhole variations in phosphate, Mn, and cc phases. In general, phases enriched in REEs (clay, phosphate, and Mn) dominate the Cretaceous, while REE-poor cc dominates the Cenozoic sediments (see Fig. 75 in Ludden, Gradstein, et al., 1990). In detail, however, distribution of the Mn and phosphate phases may be complex (Figs. 7B, 12). Nonetheless, high P2O5 contents are typical of the 450- to 600-m interval (Fig. 12). This interval is distinguished by the lowest sedimentation rates in the hole (1-2 m/m.y.), where fish debris may not be overwhelmed by other influxes. Even so, the P2O5 contents in this interval require less than 1% apatite (with about 40% P2O5), which may explain why fish debris was not identified in sediment smear slides. Because fish-bone apatite may contain on the order of 100 times the REE contents of shales (Elderfield and Pagett, 1986; Staudigel etal., 1985/86; Toyoda etal., 1990), this seemingly trivial amount becomes significant.
Although minor phases and lithologies may be important for affecting downhole variations in certain elements, the aim of this study is to characterize the average composition of Site 765 sediments. In this regard, the most volumetrically significant control on element concentrations in Site 765 sediments is dilu- tion of a shalelike phase with cc. For example, although subtle differences in REE patterns may reflect different mineral phases, all Site 765 sediments, and in fact all marine sediments, have REE patterns remarkably similar to average shale. This justifies the practice of normalizing REE to average shales. Although K2O/AI2O3 varies with clay mineralogy (Figs. 6A, 6B), the aver- age value is similar to average shale (around 0.2 for Australian shales; Taylor and McLennan, 1985). In the following section, we
25R-3-106
62R-3-80
7R-2-27
LaCe Nd SmEuGdTb YbLu LaCe Nd SmEuGdTb YbLu
Figure 10. REE patterns in selected Site 765 sediments. Gd has been interpolated between Tb and Sm to illustrate Eu anomalies better. Closed circles indicate Cenozoic sediments; open circles represent Cretaceous sediments. A. Chondrite-normalized patterns (to values in Taylor and Gorton, 1977). B. Shale-normalized patterns (to PAAS in Taylor and McLennan, 1985).
180
>. 2 oö
O
• Pacific pelagic clays O Site 765 Clays A Site 765 Cc
• .01 0.1 1
P 2 O 5 (wt%) 10
Figure 11. Ce anomalies in Site 765 clays, carbonates (>20% CaO), and Mn-rich clays. Carbonates have negative Ce anomalies; Mn-rich clays have positive anomalies. Other Site 765 clays exhibit a similar relationship between Ce anomaly and P2O5 content as recent Pacific pelagic clays (Toyoda et al., 1990).
0.0
200 400 600 800 1000
Depth (mbsf) Figure 12. Variation in P2O5 content of Site 765 sediments with depth.
discuss the provenance of this shalelike detrital phase in Site 765 sediments.
Provenance of Detrital Phase Site 765 lies oceanward of the northwestern shelf of Australia,
and thus the likely source of Site 765 sediments is northwestern Australia. A considerable data base for the sedimentary masses on the Australian continent has been developed overpast decades by researchers from the Australian National University as a way of estimating the composition of this exposed continental crust (reviewed in Taylor and McLennan, 1985). Special attention has been paid to the Archean and Archean/Proterozoic boundary sedimentary sequences; thus, a fair amount of data exists for samples from the Precambrian Pilbara Block of western Australia (see Fig. 1). A more direct source for Site 765 sediments might
be the Phanerozoic rocks of Canning Basin, but these have been ultimately derived from the Archean and Proterozoic blocks of western and central Australia, too (BMR Paleogeographic Group, 1990).
Figures 13A through 13D show AI2O3 variation diagrams for Site 765 sediments, along with sediment data from the Canning Basin (Nance and Taylor, 1976), the Pilbara Block (McLennan, 1981; McLennan et al., 1983), and the post-Archean Australian Shale (PAAS) average from Taylor and McLennan (1985). The Pilbara Block sediments include both Archean shales from the George and Whim Creek groups, as well as early Proterozoic sediments from the Hamersley Basin. The Archean shales from the George and Whim Creek groups are too depleted in Tiθ2 and Th, and too enriched in Ni and Cr, to be an appropriate source for the Site 765 sediments (Figs. 13A-13D). Furthermore, the Archean shales lack an Eu-anomaly, which is a ubiquitous feature in Site 765 REE patterns (Fig. 10A), and indeed, is the hallmark of post-Archean sediments (Taylor and McLennan, 1985). These characteristics of the George and Whim Creek shales are common to most Archean shales and preclude much of a pure Archean component in the Site 765 sediments. On the other hand, an average of three Canning Basin shales (Paleozoic) provides a good fit with the detrital end-member for Site 765 sediments (Figs. 13A-13D) and makes sense geographically (see Fig. 1) as a source for river or wind influxes to the Exmouth Plateau and Argo Abyssal Plain. The Canning Basin average is near the PAAS composite itself, which is similar to post-Archean shales else- where, owing to the remarkable mixing efficiency of sedimentary processes (Taylor and McLennan, 1985). The early Proterozoic Hamersley Basin sediments lie compositionally between the Pa- leozoic Canning Basin shales and the Archean Pilbara shales, but like the Archean shales, are too Cr-enriched and Tiθ2-depleted to be a suitable end-member for Site 765 sediments. Thus, despite significant exposures of Archean rocks in their presumed source, Site 765 sediments are compositionally like post-Archean shales. The ancient age of the source of Site 765 sediments should be reflected, however, in their compositions of Pb and Nd isotopes. Future isotopic work using these samples should help constrain the mean age of the source of Site 765 sediments.
For elements that increase down the hole (e.g., REEs, Fe), the Cenozoic values are more representative of a purely detrital com- ponent having a composition near PAAS, which is consistent with the Cenozoic section being dominated by turbidites that quantita- tively deliver continental detrital material to the marine environ- ment. The Cretaceous sediments may be enriched over PAAS because of additional hydrogenous, volcanic, hydrothermal, or phosphate contributions, as discussed above.
The proximity of active volcanoes of Java and the Lesser Sunda Islands (Bali, Lombok, Sumbawa, etc.) raises the issue of the extent to which arc andesites contributed to the youngest Site 765 sediments. Ninkovich (1979) demonstrated that Lesser Sunda ash falls extend to only 200 km south of the Sunda Trench. Site 765 lies far south of this zone, and indeed, no prominent ash fall layers were cored in the Cenozoic section. However, it is possible that a disseminated ash component contributed to Site 765 sedi- ments. Figure 14 shows the K2O and Cr contents of Site 765 sediments, modern Indonesian volcanics, and post-Archean shales. During differentiation of arc magmas, elements incompat- ible in the fractionating mineral assemblage (such as K) increase dramatically, while compatible elements (Cr) decrease dramati- cally. Thus, although arc andesites have high K2O abundances, they are strongly depleted in Cr, unlike marine clays and terrige- nous shales that are enriched in both. These systematics apply to the compatible element Ni as well, but Ni may be further enriched in marine sediments from a hydrogenous (Fe-Mn oxide) contribu- tion. The high Cr/K2θ contents throughout the Cenozoic section
181
2. CM
T iO
^ C B O * AP
15 (wt%)
20 25
Figure 13. Selected elements vs. AI2O3 in Site 765 sediments (open triangles Cretaceous, closed triangles = Cenozoic), average shales from the Phanero-
zoic Canning basin (CB), the early Proterozoic Hamersley Basin (HB), and the Archean Pilbara Block (AP), and average post-Archean Australian shale (PAAS). A. Tiθ2 vs. AI2O3. B. Sm vs. AI2O3. C. Th vs. AI2O3. D. Cr vs. AI2O3. Data are from Taylor and McLennan (1985); Nance and Taylor (1976); McLennan et al. (1983); McLennan (1981); and Tables 1 and 2.
160
120
1 2 3 4 5
K 2 O ( w t % )
Figure 14. K2O vs. Cr for Site 765 sediments (triangles), the Canning Basin average and PAAS (as in Fig. 13), and andesites from Java and Lesser Sunda (Merbabu volcano, Java, and Batur volcano, Bali). Data for andesites from Whitford (1975) and Wheller and Varne (1986). Open triangles = anomalously high Cr sediments in the 200- to 400-m interval at Site 765 (Fig. 4B).
rules out much andesitic ash in Site 765 sediments. As Site 765 approaches the Sunda Trench, however, ash may have contributed significantly to the upper sedimentary section, and subduction of this material leads to interesting speculations about the extent of cannibalism at arcs (Ben Othman et al., 1989).
CALCULATING A BULK COMPOSITION FOR SITE 765 SEDIMENTS
Estimating the bulk composition of Site 765 sediments might be as simple as averaging the analyses in Tables 1 and 2. However, this estimate is only as accurate as each sample is representative. Because cc dilution accounts for most of the geochemical varia- bility in Site 765 sediments, this estimate may be refined by taking into account cc variations. Cc contents are linked to macroscopic lithologies (foraminifer sands and nannofossil oozes); thus, pub- lished core descriptions provide, in effect, continuous downhole cc estimates. To determine the bulk composition of the site, we estimated the cc content of each 10-m interval (each core) from visual core descriptions, used our analyses to determine the com- position of the noncarbonate fraction, and then diluted this com- position by the estimated cc content. Details of these calculations are presented next.
The data reported here represent spot analyses that must be weighted by the length of the interval that they represent. For example, some sampled clay units are only centimeters thick, while some sampled nannofossil ooze units are meters long. To determine the weighting factors, the relative proportions of car- bonate-rich lithologies were estimated for each of the 103 cores from the barrel sheets in Ludden, Gradstein, et al. (1990). Because a sediment described as "nannofossil ooze" is not pure calcite, shipboard CaCθ3 analyses were used to calculate the pure cc fraction for each core. The rest of the core was considered "non- carbonate." Figure 15 presents the downhole variation in this value. This "noncarbonate" factor is simply a way to quantify and smooth lithologic variations downhole. Moreover, because most of the geochemical variability is directly linked to carbonate content, this value links lithologic and geochemical data.
The data in Table 1 were normalized to a carbonate-free, dry basis (by normalizing by a sum that does not include CaO nor the
182
100
Depth (mbsf)
800 1000
Figure 15. Downhole variation in percentage of noncarbonate per core at Site 765. This value has been estimated from CaCθ3 contents and core descriptions in Ludden, Gradstein, et al. (1990). See text for details.
LOI) and averaged for each core. These values were then "diluted" by multiplying by the noncarbonate percentages in Figure 15 to obtain average compositions for each core.
The method of weighting concentrations by the average non- carbonate content is inappropriate for those elements that are contained within carbonate (i.e., Sr), or that do not indicate carbonate dilution relationships (Ba, Mn, P, As, and U). For these elements, individual samples were assumed to be representative of each core. Where more than one analysis per core existed, values were averaged, and where no samples from a core were analyzed, values were interpolated. This method is limited by the extent to which sampling was representative of the dominant lithologies. Because Sr is contained in appreciable amounts in both carbonate (usually >IOOO ppm) and clay-rich (usually <200 ppm) units, both clay and carbonate Sr contents were estimated for each core by interpolation of actual measurements, and these values were weighted by the average carbonate content of each core.
Geochemical Logs
These average core compositions provide a smoothed down- hole data set that can be compared with geochemical logs. The natural gamma-ray tool (for K, Th, and U), aluminum activation clay tool (for Al), and the gamma-ray spectroscopy tool (for Si, Fe, Ti, and Ca) were run through casing in Hole 765D; the hole was then drilled to set a reentry cone for sampling basement (see also Pratson and Broglia, this volume). Logging through casing decreases the quality of the data because it reduces the signal and adds another factor that must be removed when processing the logs. To smooth these logging data, we applied a five-point running average.
Although logs from the gamma-ray spectroscopy tool proved to have too poor quality to use, the AI2O3 and K2O logs, acquired using two different tools, seem to agree with each other, as well as with the basic lithologic core descriptions. Figure 16 shows K2O and AI2O3 from the logging data for the hole (the upper 170 m was not logged), compared with the weighted averages calcu- lated from our analyses of the core samples. Despite simplifica- tion and smoothing of the logging data, the trends agree remarkably well with the "ground truth" analytical data. Both exhibit low values in the 200- to 400-m interval, characterized by
low clay content, a peak around 600 m, and lower values deeper in the hole. The average K2O content of the hole calculated from the logs is 1.5 wt%, while the average calculated from our sedi- ment analyses is 1.6%. The AI2O3 log in the interval between 200 and 400 m is about 6 wt% too high, and this may have resulted from a processing artifact, where the highly attenuated signal in this interval may have been overcompensated (Pratson and Bro- glia, pers. comm., 1990). However, average AI2O3 contents cal- culated for the lower 500 m agree well with the estimate from the weighted core analyses (10.6 vs. 11.1 wt%, from the logs and analyses, respectively).
Bulk Composition of Site 765 Sediments
A grand average for the entire 930-m section was calculated from the core-by-core averages, and this estimate of bulk compo- sition of the hole is presented in Table 4. Separate compositions are presented for the Cenozoic and Cretaceous sections.
Even though significant differences exist between the Ceno- zoic and Cretaceous sediments, as discussed previously, the pri- mary difference between the two sections is simply their different cc contents. For example, while Sm/Yb increases significantly from 1.9 for the Cenozoic section to 2.4 in the Cretaceous section, Sm concentration itself more than doubles as cc decreases from mean values of 60% to 10%. Thus, dilution is still the dominant signal.
A surprisingly accurate bulk composition for Site 765 sedi- ments may be obtained simply by multiplying an average shale composition (such as PAAS) by the average noncarbonate content of the hole (60%). Table 4 presents the results of these calcula- tions; most elements may be estimated within 30%. Sr can be well approximated by assuming 1500 ppm in the carbonate fraction and PAAS values in the clay fraction. A better fit can be obtained for elements such as K, Nb, Zr, Rb, and Th by taking advantage of their roughly constant and upper crustal-ratios to Al. Along with Al, all these elements are higher in the estimate based simply on PAAS and the carbonate content. By assuming PAAS ele- ment/Al2θ3 ratios and by multiplying by the mean AI2O3 content of the Site 765 sediments, better matches for these elements can be obtained. Elements having poor fits include MnO, Cu, and Ni (due to hydrogenous oxides), MgO (due to the diagenetic minerals of the 200- to 400-m interval), Ba (due to radiolarian concentra- tions) and Cs. CS/AI2O3 is virtually constant in Site 765 sedimen- ts, but much lower than PAAS.
The success of the PAAS estimate for most elements suggests that even though extra hydrogenous, volcanic, phosphate, and hydrothermal phases contribute to the Cretaceous sediments, its composition is still dominated by average shales. By assuming an average shale composition (PAAS), a relatively accurate estimate of the bulk composition of the hole can be made without relying on any chemical analyses: the carbonate dilution factor can be estimated from visual core descriptions, and average Al or K contents, which constrain crustal ratios, can be determined from geochemical logs.
SITE 765 AS A REFERENCE SECTION
Marine sediments are largely mixtures of four components: biogenic, detrital, hydrogenous, and hydrothermal (Dymond, 1981; Leinan, 1987). The hydrothermal component is only impor- tant near an active ridge-crest hydrothermal system, although disseminated components may be far reaching (1000 km, Barrett, 1987). The hydrogenous component, associated with Fe-Mn ox- ides, dominates only when the other components are absent, such as is typical for the South Pacific, much of which is below the CCD and far removed from regions of high biologic productivity and terrigenous sources. Site 765 represents the other end-mem- ber. Near a continental margin, the site is dominated by biogenic
183
| 3
*irV H H
A áU
Depth (mbsf)
Depth (mbsf)
Figure 16. Downhole logs for K2O and AI2O3 (from Pratson and Broglia, this volume) compared with weighted analyses of core samples. See text for details of weighting and data processing. Downhole K2O variations from the two methods agree remarkably well. AI2O3 content agrees well in sediments below 400 mbsf.
and detrital material. This leads directly to the success of the simple calculation based on PAAS and cc content to describe the bulk composition of Site 765. The large biogenic influx means that variations in the other components are overwhelmed by simple quantitative dilution. However, proximity to a continent also means that the detrital component is relatively constant as well, owing to the remarkable homogeneity of mature upper crustal material. Indeed, McLennan et al. (1990), in a survey of deep-sea turbidites, found that passive margin turbidites typically sample average, old, upper crustal material. These two factors, dilution because of biogenic components and average crustal detritus, should make calculating sedimentary columns adjacent to continents elsewhere just as simple. Results from Site 765 suggest that accurate estimates may be made with little analytical effort. Continental margin sections exist outboard of other arcs than Indonesia (the Antilles, Chile, Alaska, Cascades, Mediter- ranean) and thus make up a significant portion of potentially subducted material. In contrast, sedimentary sections in the mid- dle of the Pacific, such as outboard of the Tonga Arc, have been starved of biogenic and detrital components and may require a completely different way to calculate bulk compositions. Future research will be dedicated to establishing another reference sec- tion in the central South Pacific.
Although much of the variation in a sedimentary section proxi- mal to a continental margin will reduce simply to dilution of average crustal shales by biogenic material, this still leads to some interesting systematics for elements that are important tracers of crustal recycling:
1. Alkali elements (K, Rb, Cs). These elements are entirely coupled to the detrital component and will be quantitatively diluted by biogenic material. Their ratios may be similar to aver- age crustal shales, although K may vary with clay mineralogy (Fig. 6). Deeply weathered source regions may contribute more kaolinite-rich clays, leading to lower K/Al than average shales. McLennan et al. (1990) also suggested that the alkali elements may fractionate from each other during weathering, with low K/Cs contents characteristic of highly weathered sources.
2. Alkaline earth elements (Sr, Ba). In contrast, these elements have little to do with the detrital component, and thus important fractionations in alkali/alkaline earth elements occur in the marine environment. Sr substitutes for Ca in marine carbonates and may be present in concentrations up to 3000 ppm in some aragonites. Nonetheless, we have shown that the average Sr content of a sedimentary section such as that at Site 765 can be well estimated simply by assuming average cc and shale values. Although Ba
184
Table 4. Bulk composition of Site 765 sediments.
SiO2
TiO2
AI2O3
Cs Hf
Ta- W-
319 7.2
3.3 2.7 .41 2.3
2.2
.14
9.8
3.7 3.1 .67 3.5
3.8
.16
4 1 0 10
90
-25
-30
-2 31
* Samples powdered in WC excluded from estimate for these elements. PAAS from Taylor and McLennan (1985), rβnormalized to 100%.
exhibits complex behavior in the marine environment, Ba con- tents may be enormously high (10,000 ppm levels) in siliceous oozes. A rough association exists between Ba and radiolarian abundances in Site 765 sediments (Figs. 3B and 3C). This asso- ciation may result from barite nucleation on decaying siliceous skeletons in the water column (Bishop, 1988). Predominance of siliceous organisms is a characteristic of regions of high produc- tivity maintained by active upwelling, such as in equatorial re- gions. Thus, Ba has been used as a paleoproductivity indicator, as well as an equatorial reference frame for charting plate motions (Schmitz, 1987).
3. REEs. The REEs typically display post-Archean shale pat- terns and are quantitatively diluted by cc and silica. REEs may become enriched, however, by phosphate or Fe-Mn oxide phases. Positive Ce anomalies are characteristic of sediments rich in Fe-Mn oxides, while negative Ce anomalies are characteristic of sediments rich in biogenic cc or phosphate. A few arc volcanics possess negative Ce anomalies (Heming and Rankin, 1979), which might have been inherited from REE-rich phosphates.
4. High field-strength elements (HFSE; Nb, Ta, Hf, Zr). These elements, like the alkalis, are completely coupled to the detrital component, although they may become enriched in turbidite sands from heavy mineral concentrations. The high concentrations of HFSE in sediments and the notoriously low concentrations in arc volcanics provide convincing evidence against bulk assimilation of sediment in subduction zones. The transfer of material from the
subducting slab to the asthenosphere beneath arcs must be one selective to certain elements (Morris and Hart, 1983).
5. Parent/daughter ratios. Although carbonate dilution has had a dramatic effect on concentrations of elements, the ratios of several elements remain relatively constant throughout Site 765 sediments, especially in the Cenozoic sequence. Indeed, we took advantage of the constant and upper crustal ratios for several elements to refine our estimate of the bulk composition of Site 765 sediments. For example, the Sm/Nd ratio is remarkably con- stant in Site 765 sediments, even though Sm concentrations may vary by a factor of 20 because of cc dilution. Important radioactive parent/daughter ratios, however, will vary significantly with car- bonate content. The most obvious is Rb/Sr, which decreases significantly in carbonate-rich lithologies as a result of both dilution of Rb and incorporation of Sr in marine carbonates. The U/Th ratio will vary in an inverse way, because Th is quantita- tively diluted, while U is somewhat taken up in carbonates (Ben Othman et al., 1989). Thus, first-order variations in these important parent-daughter ratios may also be controlled by carbonate content.
SEDIMENT SUBDUCTION ALONG THE SUNDA TRENCH
Although not the specific aim of this study, these data for Site 765 sediments have obvious applications to sediment recycling at the Sunda Arc. Geophysical observations along the Sunda Trench are ambiguous regarding structural evidence for sediment subduc-
185
T. PLANK, J. N. LUDDEN
tion. A large gradient in sediment thickness occurs along the Sunda Trench (Fig. 17). In the northwest, south of Sumatra, there may be as much as 5 km of sediment, much of which is being accreted in the forearc (Moore et al., 1980). Farther east, south of Java, only a thin veneer of sediment approaches the trench: as little as 200 m in places (Moore et al., 1980). Available seismic data, however, show little resolvable structure within the Java Trench slope and thus leave open the question of sediment accre- tion (Curray et al., 1977).
The isotope 10Be was measured in samples from 10 volcanoes on Java and on Bali, and abundances were indistinguishable from lavas in other tectonic settings (Tera et al., 1986). A lack of 10Be enrichment in Indonesian volcanics does not prove that sediments were not subducted beneath Java. Because 10Be has such a short half-life (about 1.5 m.y.), it is present only in Neogene sediments, and the absence of 10Be in Java volcanics might simply mean that Neogene sediments were not subducted and erupted within 10 m.y. (the time over which 10Be decays). It is entirely possible that older sediments are being subducted. Whitford and Jezek (1982) suggested that sediments were incorporated in the source of Java magmas, based on the radiogenic Pb isotopic compositions and steep 207Pb/204Pb vs. 206Pb/204Pb trend defined by Java volcanics. These data have long served as one of the classic examples for sediment incorporation, even though no sediment data were avail- able at the time Whitford and Jezek (1982) formed their model. However, new Pb isotopic data for piston core samples of sedi-
ments outboard of Java overlap with the volcanics (Ben Othman et al., 1989) and provide new support for the original interpreta- tion. Thus, the most compelling argument for subduction comes from Pb isotopic data.
Some data exist with which to extrapolate the lithologic and geochemical stratigraphies developed at Site 765 regionally along the Sunda Arc. Several holes that penetrated basement were drilled during DSDP Legs 22 and 27, and although recovery was poor (often less than 20%), data from these holes do provide a means of estimating the lithologic sections around the Sunda Trench (see Fig. 17). These holes include two around the Australian margin (Sites 261 and 260), two farther west in the Wharton Basin (Sites 212 and 213), and one just south of the Sunda Trench (Site 211). From descriptions in the Initial Reports volumes of these cruises, it appears that calcareous turbidites from the Australian margin extend as far north as Site 261 and as far west as Site 212 in the Wharton Basin. Site 213 contains little carbonate, but has accumu- lated siliceous oozes since the Miocene. An extensive section of Site 211 is composed of clastic turbidites of the Nicobar Fan.
A small amount of geochemical data has been published for sediments from Sites 261 and 260 (largely major elements; Cook, 1974), Sites 211 through 213 (transition metals; Pimm, 1974), and for a few piston cores outboard of the Sunda Trench (trace ele- ments and isotopes; Ben Othman et al., 1989; see Fig. 17). Be- cause no shared set of elements exists, one finds it difficult to draw many regional conclusions. Future research will include isotopic
10°N
-\ 10°S
80°E 100 110° 120° 130c 40°
Figure 17. Location map of other DSDP and ODP holes (closed circles) throughout the region outboard of the Sunda Trench. Water depth in meters. Open circles are piston cores V34-47, V34-45, V28-341, and V28-343 from Ben Othman et al. (1989) that are discussed in the text (other piston cores from Ben Othman et al., 1989, are on the north side of the Sunda Trench and are not discussed here).
186
GEOCHEMISTRY REFERENCE SECTION FOR SEDIMENT RECYCLING
analyses of Site 765 sediments, as well as analyses of major and trace elements in the piston core samples for which Ben Othman et al. (1989) analyzed isotopes. These data will allow for more confident extrapolation of information along the entire Sunda Arc. However, the following preliminary observations can be made. Sites 260 and 261 along the Australian margin are similar to Site 765 in that the Cretaceous section is dominated by clays, the Cenozoic section contains calcareous turbidites, and the detri- tal end-member, at least based on the K2θ/Tiθ2 ratios (Fig. 18), appears similar to PAAS. Although one would wish to analyze more clay samples for more elements, calculating the bulk com- position of these two sites may be as simple as estimating the cc and Al contents, as we demonstrated for Site 765 sediments.
MnO-rich clays may be more common from the Wharton Basin sites than from those sites nearer the Australian margin (Fig. 19).
O
A
Cfc>A D PAAS U O A
A .
0.0 0.2 0.4 0.6 0.8 1.0 1.2
TiO2 wt% Figure 18. K2O vs. Tiθ2 for sediments around the Argo Abyssal Plain. Data for DSDP Sites 260 and 261 from Cook (1974).
2 -
O
1 "
A • •
ML
8 10 12 FeO (wt%)
Figure 19. FeO* vs. MnO for sediments from DSDP Sites 212 and 213 in or near the Wharton Basin and Site 765. Data for Sites 212 and 213 from Pimm (1974).
Although no higher in concentration than the most manganiferous sediments from Site 765, almost all of the samples analyzed from Sites 212 and 213 are rich in MnO, perhaps reflecting a. greater hydrogenous/detrital fraction in the clays. This interpretation is consistent with these sections being farther from the Australian continent and their detrital sources. The isotopic compositions of the piston core samples near the Australian margin are distinctive of old cratonic material (i.e., high ^ P b / ^ P b , low 1 4 3 Nd/ 1 4 4 Nd), while those west of the Wharton Basin are distinctly less enriched (Fig. 20), perhaps reflecting influx of younger material from the Nicobar Fan. The Wharton Basin sample has higher Lu contents than any sample from Site 765, while other piston core samples as far west as the Ninetyeast Ridge and as far east as the Banda Islands overlap completely with Site 765 sediments (Fig. 21).
Existing data are patchy, but suggest it may be possible to extrapolate at least certain aspects of Site 765 sediments across the entire basin from the Banda Islands to Ninetyeast Ridge. Australian continental detritus is most likely an important com- ponent of sediments at least out into the Wharton Basin. Key analyses of other important components, such as Mn-rich clays of the Wharton Basin, clastic material fed from the Nicobar Fan, and ash from the active arc, coupled with careful consideration of the lithologic characteristics of the DSDP holes (e.g., carbonate vs. clay fractions), might yield good first-order estimates of elemen- tal fluxes into the Sunda Trench.
CONCLUSIONS
1. The dominant signal in the geochemical variability of Site 765 sediments is dilution of a detrital component by biogenic calcium carbonate and silica. This dilution leads to enormous variations in the concentrations of most elements. Dilution from carbonate or silica may be a long-lived feature of the sedimentary column, even if subducted to great depths because of the relatively high temperatures required for decarbonation reactions (Gill, 1981; Abbott and Lyle, 1984).
Q_ ^ J •
o CVJ
JO Q_
0 Wharton Basin
Ninety East Ridge
N Australia margin -• 80 90 100 110 120 130 140
Longitude (°E)
Figure 20. Pb and Nd isotopic compositions for surface samples outboard of the Sunda Trench (from Ben Othman et al., 1989).
187
α. Q.
1 0 -
0.0 0.2 0.8 1.00.4 0.6
Lu (ppm) Figure 21. Sm and Lu concentrations for Site 765 sediments and the four piston
core samples shown in Figure 20 from the Banda region (BAN), Argo Abyssal
Plain (AAP), Ninetyeast Ridge (90E), and Wharton Basin (WB).
2. Significant differences occur between Cenozoic and Creta- ceous clays, involving increases in Fe, Ti, Mn, Ba, REEs, changes in REE patterns, and development of positive Ce anomalies down the section. These variations may result from increases in hy- drogenous, hydrothermal, phosphate, and/or volcanic phases dur- ing the Cretaceous. Although important for identifying changes in the provenance, sediment supply, or rate with time, these phases led to small deviations from the average composition of the hole, which is dominated by dilution of an average shale composition.
3. The K2O and AI2O3 downhole logs correspond well with "ground truth" chemical analyses. The bulk composition of the hole can be calculated by using the visual core descriptions to weight the individual analyses over appropriate intervals. This composition, however, agrees remarkably well (to 30% for most elements) with an estimate based simply on average Australian shales and the average cc and AI2O3 contents of the hole. This result suggests that estimating other sections dominated by car- bonate and continental detritus may require minimal analytical effort. Ideally, only core descriptions and logging data should provide estimates of cc and Al contents that are accurate enough to characterize a site.
4. Although Site 765 is an important reference section for sedimentary columns along the Sunda Arc, our results are more general. Site 765 sediments are well described by dilution of average shale by biogenic phases, and because average shale compositions are remarkably similar around the world, results from Site 765 are general, not restricted to provenances or pro- cesses about the Indonesian region. Site 765 should thus serve as a useful reference for calculating other continental margin sec- tions approaching trenches around the world (e.g., the Antilles, Americas, Mediterranean). A recent study by Hay et al. (1988) about global distribution of sediments on the ocean floor esti- mated that roughly 40% of ocean sediments is calcium carbonate and roughly 45% is terrigenous detrital material. Therefore, Site 765 sediments, dominated by cc and terrigenous detritus, is repre- sentative of a large part of the global marine sedimentary reservoir.
ACKNOWLEDGMENTS
We thank Ginger Eberhart, Gilles Gauthier, and Bettina Domeyer for technical assistance with the DCP, INAA, and XRF analyses, respectively. Scott McLennan kindly provided unpub-
lished data and a preprint. We thank Charlie Langmuir for helpful discussions and support, and S. M. McLennan and C. R. Czerna for useful reviews. Mitch Lyle and John Compton are especially appreciated for making themselves available for an onslaught of questions. T. Plank gratefully acknowledges financial support from USSAC and from a JOI/USSAC Graduate Fellowship.
REFERENCES
Abbott, D., and Lyle, M., 1984. Age of oceanic plates at subduction and volatile recycling. Geophys. Res. Lett., 11:951-954.
Aplin, A. C , 1984. Rare earth element geochemistry of Central Pacific ferromanganese encrustations. Earth Planet. Sci. Lett., 71:13-22.
Barrett, T., Taylor, P. N., and Lugowski, J., 1987. Metalliferous sediments from DSDP Leg 92: the East Pacific Rise transect. Geochim. Cosmo- chim. Acta, 51:2241-2253.
Ben Othman, D., White, W. M., and Patchett, J., 1989. The geochemistry of marine sediments, island arc magma genesis and crust-mantle recycling. Earth Planet. Sci. Lett., 94:1-21.
Bishop, J.K.B., 1988. The barite-opal organic carbon association in oceanic particulate matter. Nature, 332:341-343.
BMR Paleogeographic Group, 1990. Australia: Evolution of a Conti- nent: Canberra (Australian Govt. Publ. Serv.).
Cook, P. J., 1974. Major and trace element geochemistry of sediments from DSDP, Leg 27, Sites 259-263, eastern Indian Ocean. In Veevers, J. J., Heirtzler, J. R., et al., Init. Repts. DSDP, 27: Washington (U.S. Govt. Printing Office), 481^98.
Curray, J. R., Shor, G. G., Raitt, R. W., and Henry, M., 1977. Seismic refraction and reflection studies of crustal structure of the Eastern Sunda and Western Banda arcs. /. Geophys. Res., 82:2479-2489.
Dymond, J., 1981. The geochemistry of Nazca Plate sediments: an evalu- ation of hydrothermal, biogenic, detrital, and hydrogenous sources. In Kulm, L. D., Dymond, J., Dasch, D. M., and Hussong, D. M. (Eds.), Nazca Plate: Crustal Formation and Andean Convergence. Geol. Soc. Am. Mem., 154:133-174.
Elderfield, H., and Greaves, M. J., 1981. Negative cerium anomalies in the rare earth element patterns of oceanic ferromanganese nodules. Earth Planet. Sci. Lett., 55:163-170.
Elderfield, H., and Pagett, R., 1986. Rare earth elements in ichtyholiths: variations with redox conditions and depositional environment. Sci. Total Environ., 49:175-197.
Francis, D., and Ludden, J. N., 1990. The mantle source for olivine nephelinite, basanite and alkali olivine basalt at Fort Selkirk, Yukon, Canada./. Petrol., 31:371^00.
Gill, J. B., 1981. Orogenic Andesites and Plate Tectonics: New York (Springer-Verlag).
Govindaraju, K., 1989. 1989 compilation of working values and sample description for 272 geostandards. Spec. Iss. Geostds. Newsl., 13:1— 113.
Hay, W. M., Sloan, J. L., and Wold, C. N., 1988. Mass/age distribution and composition of sediments on the ocean floor and the global rate of sediment subduction./. Geophys. Res., 93:14933-14940.
Heming, R. F., and Rankin, P. C , 1979. Ce-anomalous lavas from Rabaul caldera, Papua New Guinea. Geochim. Cosmochim. Acta, 43:1351- 1355.
Hussong, D. M., Uyeda, S., et al., 1981. /n/r. Repts. DSDP, 60: Washing- ton (U.S. Govt. Printing Office).
Karig, D. E., and Kay, R. W., 1981. Fate of sediments on the descending plate at convergent margins. Philos. Trans. R. Soc. London, 301:233- 251.
Klein, E. M., Langmuir, C. H., and Staudigel, H. S., 1991. Geochemistry of basalts from the SE Indian Ridge, 115°-138°. /. Geophys. Res., 96:2089-2108.
Leinen, M., 1987. The origin of paleochemical signatures in North Pacific pelagic clays: partitioning experiments. Geochim. Cosmochim. Acta, 51:305-319.
Ludden, J. N., Gradstein, F. M., et al., 1990. Proc. ODP, Init. Repts., 123: College Station, TX (Ocean Drilling Program).
McLennan, S. M., 1981. Trace element geochemistry of sedimentary rocks: implications for the composition and evolution of the continen- tal crust [Ph.D. dissert.]. Australian National Univ., Canberra.
188
GEOCHEMISTRY REFERENCE SECTION FOR SEDIMENT RECYCLING
McLennan, S. M., Taylor, S. R., and Eriksson, K. A., 1983. Geochemistry of Archean shales from the Pilbara Supergroup, Western Australia. Geochim. Cosmochim. Acta, 47:1211-1222.
McLennan, S. M., Taylor, S. R., McCulloch, M. T., and Maynard, J. B., 1990. Geochemical and Nd-Sr isotopic composition of deep-sea tur- bidites: crustal evolution and plate tectonic associations. Geochim. Cosmochim. Acta, 54:2015-2050.
Moore, G. F., Curray, J. R., Moore, D. G., and Karig, D. E., 1980. Variations in geologic structure along the Sunda fore arc, northeastern Indian Ocean. In Hayes, D. E. (Ed.), The Tectonic and Geologic Evolution of Southeast Asian Seas and Islands. Am. Geophys. Union, 145-160.
Morris, J. D., and Hart, S. R., 1983. Isotopic and incompatible element constraints on the genesis of island arc volcanics from Cold Bay and Amak islands, Aleutians, and implications for mantle structure. Geo- chim. Cosmochim. Ada, 47:2015-2030.
Nance, W. B., and Taylor, S. R., 1976. Rare earth element patterns and crustal evolution. I. Australian post-Archean sedimentary rocks. Geo- chim. Cosmochim. Ada, 40:1539-1551.
Ninkovich, D., 1979. Distribution, age and chemical composition of tephra layers in deep-sea sediments off western Indonesia. / . Vol- canol. Geotherm. Res., 5:67-86.
Palmer, M. R., 1985. Rare earth elements in foraminifera tests. Earth Planet. Sci. Lett., 73:285-298.
Pimm, A. C , 1974. Mineralization and trace element variaitons in deep- sea pelagic sediments of the Wharton basin, Indian Ocean. In von der Borch, C. C , Sclater, J. G., et al., Init. Repts. DSDP, 22: Washington (U.S. Govt. Printing Office), 469-476.
Schmitz, B., 1987. Barium, equatorial high productivity, and the north- ward wandering of the Indian continent. Paleoceanography, 2:63-77.
Schweller, W. J., Kulm, L. D., and Prince, R. A., 1981. Tectonics, structure, and sedimentary framework of the Peru-Chile Trench. Mem. Geol. Soc. Am., 154:323-349.
Staudigel, H., Doyle, P., and Zindler, A., 1985. Sr and Nd isotopic systematics in fish teeth. Earth Planet. Sci. Lett., 76:45-46.
Taylor, S. R., and Gorton, M. P., 1977. Geochemical application of spark source mass spectrometry. HI. Element sensitivity, precision and accuracy. Geochim. Cosmochim. Acta, 41:1375-1380.
Taylor, S. R., and McLennan, S. M., 1985. The Continental Crust: Its Composition and Evolution: Oxford (Blackwell Scientific).
Tera, F., Brown, L., Morris, J., Sacks, I. S., Klein, J., and Middleton, R., 1986. Sediment incorporation in island-arc magmas: inferences from lOBe. Geochim. Cosmochim. Acta, 50:535-550.
Toyoda, K., Nakamura, Y., and Masuda, A., 1990. Rare earth elements of Pacific pelagic sediments. Geochim. Cosmochim. Acta, 54:1093- 1103.
von Huene, R., 1986. To accrete or not to accrete, that is the question. Geol. Rundsch., 75:1-15.
Warsi, W.E.K., Hilde, T.W.C., and Searle, R. C , 1983. Convergence structures of the Peru Trench between 10°S and 14°S. In Hilde, T.W.C., and Uyeda, S. (Eds.), Convergence andSubduction. Tectono- physics, 99:313-329.
Watkins, J. S., Moore, J. C , et al., 1982. Init. Repts. DSDP, 66: Wash- ington (U.S. Govt. Printing Office).
Westbrook, G. K., Ladd, J. W., Buhl, P., Bangs, N., and Tiley, G. J., 1988. Cross section of an accretionary wedge: Barbados Ridge complex. Geology, 16:631-635.
Wheller, G. E., and Varne, R., 1986. Genesis of dacitic magmatism at Batur volcano, Bali, Indonesia: implications for origins of stratovol- cano calderas. J. Volcanol. Geotherm. Res., 28:363-378.
Whitford, D. J., 1975. Geochemistry and petrology of volcanic rocks from the Sunda arc, Indonesia [Ph.D. dissert.]. Australian National Univ., Canberra.
Whitford, D. J., and Jezek, P. A., 1982. Isotopic constraints on the role of subducted sialic material in Indonesian island-arc magmatism. Geol. Soc. Am. Bull., 93:504-513.
Date of initial receipt: 19 July 1990 Date of acceptance: 16 August 1991 Ms 123B-158
189